A Survey of Content-Based Video Segmentation on the Compressed Video Stream

An overview of existing techniques for video segmentation in the compressed domain, including their relative advantages and limitations is presented. According to the main contents of information used, these methods for video segmentation were classified into five types, namely：using DCT coefficients, DC terms, motion vectors（MV）, DC terms and MV, macro block coding and MV. In addition, recent research status in this area and the existing problems were also described. Finally, the research prospect was discussed.